Heat Shock Transcription Factor 1 Inhibits Expression of IL-6 through Activating Transcription Factor 3

Inflammatory cytokines, such as IL-1 and -6 and TNF-α, are produced in response to bacterial infections and disease and elicit the febrile response, a complex physiological reaction to disease, including a cytokine-mediated increase in body temperature and the activation of inflammatory systems (1, 2). Fever plays beneficial roles in the clinical prognosis of disease (3, 4). Experimentally, the pretreatment of animals with heat shock also increased survival in a LPS-injected endotoxic model (5, 6). These beneficial roles of fever are mediated partly by a strengthening of immune surveillance through lymphocyte trafficking (7), as well as by the suppression of pyrogenic and inflammatory cytokines, such as IL-1 and TNF-α, in the cell (8–10) and body (6, 11). However, it is unclear whether hyperthermia inhibits the expression of IL-6 as it affects IL-6 and TNF-α expression differently in macrophages (10).

One major adaptive response to high temperature stress in all living organisms is the production of heat shock proteins (HSPs) that assist protein folding and inhibit protein denaturation (12). This response is regulated mainly at the level of transcription by heat shock transcription factor 1 (HSF1) (13–15). HSF1 protects cells from exposure to extreme temperatures, by inducing the expression of HSP (16–19), and from various pathophysiological conditions, such as age-related neurodegenerative diseases (20–23). Because HSF1 can sense even a mild increase in temperature, it has been suggested to play roles in the regulation of genes for inflammatory cytokines (24). In fact, analysis in vivo using an LPS-injected endotoxic model showed that mortality is high in mice lacking HSF1, which is associated with an increase in the serum TNF-α level (25). Furthermore, HSF1 inhibits the expression of cytokines by binding directly to the TNF-α promoter (26) or by physically interacting with NF–IL-6, an activator for IL-1β (27). However, we do not know the involvement of heat shock or HSF1 in the expression of hundreds of inflammatory genes, including IL-6, or the mechanisms by which HSF1 suppresses the expression of these genes.

We previously found that HSF1 directly binds to the IL-6 promoter to open the chromatin structure in the absence of heat shock (28, 29). Thus, HSF1 is required for the LPS-mediated maximal expression of IL-6 in macrophages and spleen cells. In this study, we extended our previous findings and examined the molecular links between heat shock and inflammatory responses. Unexpectedly, we found that activation of HSF1 inhibited LPS-induced IL-6 expression in macrophages and mouse embryonic fibroblast (MEF) cells. The activated HSF1 induced the expression of activating transcription factor (ATF) 3, a negative regulator of the IL-6 gene (30). Furthermore, the HSF1-ATF3 pathway was required for heat-mediated suppression of inflammatory genes, including IL-6, in cultured cells and in mice.

Primary cultures of wild-type, HSF1^−/− (19) and ATF3^−/− (31) MEFs were prepared and maintained at 37°C in 5% CO₂ in DMEM containing 10% FBS. Macrophages were collected as adherent peritoneal cells (28). Cells were treated with heat shock at 42°C for 1 h and then with LPS Escherichia coli 0127:B8 (Sigma-Aldrich, St. Louis, MO) (1 μg/ml) for specific periods. IL-6 levels in culture media were determined using an ELISA kit (Invitrogen, Carlsbad, CA).

Total RNA was isolated from MEF cells or tissues using TRIzol (Invitrogen); a Northern blot analysis was performed, as described previously, using [³²P]-labeled cDNA probes for mouse IL-6, HSP70-1, and β-actin (28) and a [³²P]-labeled BamHI/EcoRI fragment of pcDNA3.1-mATF3 (29). An autoradiographic image on Fuji medical X-ray film (Super HR-HA, Fujifilm, Tokyo, Japan) was scanned, and levels of mRNA were estimated using the NIH Image program. RT-PCR analysis of mRNA levels was performed essentially as described previously (28), using specific primers for inflammatory genes (Supplemental Table I). The amplified DNA was stained with ethidium bromide and photographed using an Epi-Light UV FA1100 system (Aisin Cosmos R&D, Kariya, Japan).

Cell extracts were prepared from MEF cells and peritoneal macrophages in NP-40 lysis buffer (29) and subjected to Western blotting using a rabbit antiserum for HSF1 (α-mHSF1j) (32), a rabbit polyclonal IgG for ATF3 (C-19; Santa Cruz Biotechnology, Santa Cruz, CA), a mouse monoclonal IgG for HSP70 (W27; Santa Cruz), GFP (Nacalai Tesque, Kyoto, Japan), or β-actin (AC-15; Sigma-Aldrich). Signals were detected using ECL Western blotting detection reagents (General Electric, Fairfield, CT).

Primary cultures of MEF cells were plated on 60-mm dishes and incubated in 2 ml serum-free DMEM containing adenovirus (2 × 10⁷ PFU/ml) for GFP, human HSF1 (hHSF1), an active hHSF1 (hHSF1ΔRDT), or an inactive hHSF1 (hHSF1R71G or hHSF1ΔAB) (29) for 2 h. After adding 2 ml DMEM containing 20% FBS, cells were maintained for 48 h and then treated with LPS for 2 or 4 h. Cell extracts were prepared in NP-40 lysis buffer for Western blot analysis, or total RNA was isolated from the cells for Northern blot analysis. To restore ATF3 expression, ATF3-null MEF cells were infected with an adenovirus expressing ATF3 (29), as described above, and treated with heat shock at 42°C and/or LPS (1 μg/ml).

A DNA fragment of region 1 (−418 to +69) relative to the transcription start site in the ATF3 promoter was isolated by PCR using genomic DNA of mouse TT2 ES cells as a template. Primers used were ATF3 pro5-2: 5′-CTA AGC TTT TAG AGA AAG GTC GTG CCC GC -3′ and ATF3 pro3: 5′-GTG TCG ACA GCG CGT TGC ATC ACC CCT TT-3′ (boldface nucleotides are HindIII and SalI recognition sites, respectively). The fragments were inserted upstream of the HSV-thymidine kinase promoter of ptk-galp3-luc at the HindIII/SalI site to generate a pATF3-luc reporter plasmid. We deleted a region containing putative heat shock elements (HSEs) HSE2 to HSE4 (−390 to −356) (pATF3-luc-m1) or a region containing putative HSE5 to HSE9 (−305 to −249) (pATF3-luc-m2). A DNA fragment of the HSP70 promoter (−517 to +21) was amplified by PCR and inserted into ptk-galp3-luc at the SalI/BamHI site to generate the pHSP70-luc reporter plasmid. HEK293 cells were transfected with 2 μg the reporter plasmid and 2 μg pact-LacZ as an internal control by the calcium phosphate method. At 4 h after the transfection, cells were washed with PBS and incubated for an additional 48 h in normal medium. After heat shock at 42°C for 1 h and recovery at 37°C for 8 h, cell extracts were prepared, and luciferase and β-galactosidase activities were examined. To examine the combined effects of LPS treatment and physiological heat shock, HEK 293 cells stably expressing TLR 4 (a gift from Dr. K. Miyake, the Institute of Medical Science, the University of Tokyo) were transfected with plasmids as described above, heat-shocked at 40°C for 1 h, and allowed to recover for 4 h in the absence or presence of LPS.

Total RNA was prepared from two MEF cell cultures derived from two independent embryos at embryonic day 15.5 (ICR mice). Cells were maintained at 37°C (control), treated with LPS for 4 h, or treated with LPS for 4 h after pretreatment with heat shock at 42°C for 1 h. Gene expression was analyzed using a GeneChip Mouse Genome 430 2.0 Array (Affymetrix, Santa Clara, CA) containing ∼45,000 probe sets. Target cRNA was prepared from 1 μg the total RNA with a One-Cycle cDNA Synthesis Kit and 3′-Amplification Reagents for IVT Labeling (Affymetrix). Hybridization to the microarrays, washing and staining, and scanning were performed according to the manufacturer’s instructions. The scanned image data were processed using the GeneChip Operating Software version 1.4 (Affymetrix), and the change (increase or decrease) for each probe set was evaluated by a Comparison Analysis with the software.

Primary MEF cells were exposed to heat stress at 42°C for 1 h, and a chromatin immunoprecipitation (ChIP) assay was performed using a kit (Upstate Biotechnology, Lake Placid, NY), according to the manufacturer’s instructions (31). An antiserum for HSF1 (α-cHSF1c) was used to precipitate complexes of HSF1 and DNA fragments (29). ChIP-enriched DNA was amplified using primers corresponding to region 1 in the ATF3 promoter (Supplemental Fig. 2) or primers corresponding to the IL-6 promoter (−565 to −827) (29). A rabbit polyclonal IgG for ATF3 (C-19) was used to detect the binding of ATF3 to the IL-6 promoter (−230 to +31) in vivo (29). The amplified DNA was stained with ethidium bromide, photographed, scanned, and quantified by NIH Image.

HSF1^−/− (19), ATF3^−/− (32), and IL-6^−/− mice (The Jackson Laboratory, Bar Harbor, ME) mice were maintained through crosses with ICR mice. The animals were kept at 24°C with lights on from 08:00 to 20:00. To analyze responses to LPS treatment, 12-wk-old mice were injected i.p. with PBS as a control or with 40 μg/kg LPS and 400 mg/kg D-(+)-galactosamine hydrochloride (GalN) (Sigma-Aldrich) (33) at zero time (10:00 am). Mice were killed by cervical translocation, and tissues were removed for Northern blot analysis. Serum IL-6 levels were determined using an ELISA kit, as described previously (28). All experimental protocols were reviewed by the Committee for Ethics on Animal Experiments of Yamaguchi University Graduate School of Medicine.

Core body temperature was monitored using telemetry. We used a commercially available data-acquisition system composed of an implantable telemetry transmitter (PhysioTel TA10ETA-F20, Data Science International, St. Paul, MN) and a receiver (PhysioTel Receiver RPC-1) placed under the cage of each animal and connected to a data-acquisition matrix. Ten days before the experiments, 6-wk-old male mice were anesthetized with an i.p. administration of Nembutal (500 mg/kg) (Dainippon Pharmaceutical, Osaka, Japan) and implanted with a transmitter. Mice were injected i.p. with LPS at 50 or 150 μg/kg (34) at 10:00 am. Body temperature was recorded at 10-min intervals, beginning ≥24 h before, and continuing for ≥24 h after the injection of LPS.

Values were tested for significance with the Student t test or two-way ANOVA. p < 0.05 was considered significant.

We found that the LPS-induced expression of IL-6 was suppressed by pretreatment with heat shock at 42°C for 1 h like that of TNF-α and IL-1β in macrophages (Fig. 1A). LPS-induced IL-6 mRNA expression and protein production were also inhibited by the same heat pretreatment in MEFs for 12 and 24 h, respectively (Fig. 1B, 1C). We next examined the involvement of HSF1 in the heat-mediated suppression of LPS-induced IL-6 expression. Overexpression of the wild-type or actively mutated HSF1ΔRDT suppressed the LPS-induced expression of IL-6 mRNA, similar to the pretreatment with heat shock, whereas overexpression of two nonfunctional mutants, HSF1R71G and HSF1ΔAB, had no effect (Fig. 1D). These results demonstrate that LPS-induced IL-6 expression is suppressed through the activation of HSF1.

We previously identified genes whose expression increased after heat shock in wild-type and HSF1-null cells (28). Among them, ATF3 had much higher mRNA and protein levels after heat shock at 42°C in wild-type cells than in HSF1-null cells (Fig. 2A, Supplemental Fig. 1). Maximal ATF3 expression was induced in cells treated with severe heat shock at 42°C, irrespective of LPS treatment (data not shown). However, mRNA and protein levels were significantly elevated in cells treated with LPS and heat shock at 40°C compared with levels in cells treated with LPS or heat shock alone (Fig. 2B, 2C). Furthermore, overexpression of wild-type or actively mutated HSF1 promoted LPS-induced expression of ATF3 mRNA similar to the pretreatment with heat shock, whereas the overexpression of nonfunctional HSF1 mutants had no effect (Fig. 2D), indicating a synergistic induction of ATF3 expression by treatment with LPS and heat shock.

There are ≥17 conserved HSE-like sequences in the mouse and human ATF3 promoter (up to −500 bp) (Supplemental Fig. 2). To identify the HSF1-binding region, we performed a ChIP assay and found that HSF1 binds to at least region d (−396 to −299) of the ATF3 promoter (Fig. 2E), which corresponds to three overlapping HSEs (HSE2–4). Reporter activity was induced after heat shock at 42°C for 1 h in the presence of HSE2–4, but not when HSE2–4 was deleted (Fig. 2F). Furthermore, HSE2–4 is necessary for the synergistic effect on reporter activity of LPS treatment and heat shock at 40°C (Fig. 2G).

We next examined the expression of IL-6 in ATF3-null MEF cells (32). Levels of IL-6 mRNA were slightly higher up to 6 h after LPS treatment in ATF3-null cells than in wild-type cells (30), and pretreatment at 42°C for 1 h resulted in a marked difference in the levels of IL-6 mRNA between wild-type and ATF3-null cells (Fig. 3A). Surprisingly, pretreatment of ATF3-null cells with heat shock suppressed neither LPS-induced IL-6 mRNA expression nor LPS-induced IL-6 production in the medium (Fig. 3B, 3C). Re-expression of ATF3 in ATF3-null cells partially suppressed the LPS-induced IL-6 mRNA expression and fully reversed the suppressive effects of heat pretreatment on the expression of IL-6 mRNA and protein. Endogenous ATF3 induced to express by heat shock or heat shock plus LPS treatment bound to the IL-6 promoter in vivo at 4 h after the heat treatment (Fig. 3D). However, the overexpression of ATF3 was not sufficient for the binding, and heat shock promoted the binding to the IL-6 promoter, probably because the nuclear translocation of ATF3 is accelerated by heat shock (Supplemental Fig. 3). We next examined the expression of ATF3 and IL-6 in peritoneal macrophages and found that pretreatment at 42°C for 1 h markedly promoted LPS-induced ATF3 expression (Fig. 3E) and that pretreatment with heat shock barely suppressed LPS-induced IL-6 production in macrophages lacking ATF3 (Fig. 3F). These results clearly indicate that ATF3 is required for the suppression of LPS-induced IL-6 expression by heat shock and exclude the possibility that the heat pretreatment impaired gene expression or signal transduction in a nonspecific manner.

HSF1 directly or indirectly inhibits the expression of TNF-α and IL-1β (26, 27). Therefore, we next examined whether the HSF1-ATF3 pathway is involved in heat-mediated suppression of inflammatory gene expression. We performed a comprehensive analysis of gene expression using a DNA microarray to identify inflammatory genes whose expression is suppressed by heat shock. We found 100 genes that showed a >3-fold increase in expression on LPS treatment (Fig. 4A). Remarkably, 86 genes (86%) exhibited a <2-fold increase after combined treatments with heat shock and LPS in MEF1 and MEF2 cells. Although IL-6 was not one of these 86 genes, its expression increased 6.8-fold in MEF1 cells treated with LPS but only 2.6-fold in the cells exposed to heat shock and LPS (GEO record GSE16266). Among 40 genes whose expression increased >8-fold in the cell culture treated with LPS but <2-fold in response to heat shock and LPS, 24 genes were confirmed by RT-PCR to be markedly suppressed in their LPS-induced expression on pretreatment with heat shock (Fig. 4B). The LPS-induced mRNA expression of 16 (67%) of these genes was not suppressed in ATF3-null cells, even in the presence of heat pretreatment like that of IL-6, and re-expression of ATF3 in ATF3-null cells partially suppressed the induction (Fig. 3B). These results demonstrate that the expression of the 16 genes is ATF3 dependent. In contrast, a deficiency of ATF3 only slightly affected the LPS-induced expression of the other eight genes (33%), including TNFα and IL-1β, and re-expression of ATF3 in ATF3-null cells had no effect on the LPS-induced expression of these genes (Fig. 3B).

Furthermore, we found that HSF1 is required for the maximal mRNA expression of 8 ATF3-dependent genes (33%) (Fig. 4B, 4C). Pretreatment of HSF1-null MEF cells with trichostatin A, a histone deacetylase inhibitor, restored the LPS-induced expressions of NOS2, ICAM1, and IL-6 mRNAs (Fig. 4D). Because the promoters of NOS2 and ICAM1 have been characterized (35, 36), we analyzed them and found that HSF1 bound directly to them (Fig. 4E, Supplemental Fig. 4), promoting the acetylation and inhibiting the methylation of histone H3K9 (Fig. 4F). These results indicate that HSF1 regulates the expression of a set of inflammatory genes indirectly, by activating ATF3, and directly, by binding to the promoters, which facilitates the binding of ATF3 to the promoters through chromatin opening.

We next examined the impact of HSF1 on IL-6 expression in mice injected i.p. with LPS. The IL-6 mRNA level in each tissue reached a peak at 2 h after the injection in wild-type mice and then declined quickly (Supplemental Fig. 5). Remarkably, these levels were much higher in the heart, liver, and kidney of HSF1-null mice than in the same tissues of wild-type mice, whereas they were only slightly higher in the spleen, probably as a result of the effect of HSF1 on chromatin (28, 29). As a result of elevated levels of IL-6 mRNA in most tissues of HSF1-null mice, serum IL-6 concentrations were extremely high in HSF1-null mice, like in ATF3-null mice (30), compared with those in wild-type mice (Fig. 5A).

The major systemic effects of an elevated serum IL-6 level are an acute-phase response and the febrile response (1, 37), both of which are impaired in IL-6–null mice (38, 39). We found the mRNAs of acute-phase proteins, α₂-macroglobulin and serum amyloid A1 and A2, to be expressed more in the liver of HSF1-null mice (Supplemental Fig. 6). Furthermore, body temperatures of HSF1-null and ATF3-null mice were significantly higher after the injection of a low dose of LPS (50 μg/kg) compared with wild-type mice, whereas body temperatures of IL-6–null mice did not increase after the injection of a low dose of LPS (150 μg/kg) (Fig. 5B). These results demonstrate that HSF1 and ATF3 suppress febrile responses in mice injected with LPS, at least by inhibiting the expression of IL-6.

Consistent with the finding that the body temperature of IL-6-null mice did not increase, levels of HSF1 bound to the ATF3 promoter were much lower at 2 and 4 h after the LPS injection in IL-6-null mice than in wild-type mice (Fig. 6A). Furthermore, levels of ATF3 translocated into the nucleus in the liver were reduced (Fig. 6B), and levels of ATF3 bound to the IL-6 promoter were also much lower (Fig. 6C). These results indicate that an IL-6–mediated increase in body temperature promotes the activation of HSF1 and ATF3.

We identified the ATF3 gene as a major target of HSF1. ATF3 is a member of the ATF/CREB family of transcription factors and was induced in response to various types of stress, including serum stimulation, inhibition of protein synthesis, ionizing radiation, and UV radiation (40), in part through the ERK, JNK, and p38 pathways (41, 42), and was involved in cell growth and survival (31, 43). However, ATF3 played no role in the survival of MEF cells at high temperature or in cell growth after a single exposure to high temperature, whereas HSF1 did (Supplemental Fig. 7). More recently, it was shown that ATF3 inhibits inflammation in various tissues (30, 44–46). Surprisingly, in this study, we demonstrated that ATF3 is required for heat-mediated suppression of the LPS-induced expression of IL-6 and many other inflammatory genes.

This finding led us to conclude that heat shock inhibits LPS-induced IL-6 expression in mouse cells. Previous reports showed that IL-6 expression was not inhibited by hyperthermia in human macrophages treated with LPS (10) and that hyperthermia promoted IL-6 production in human enterocytes (47). Furthermore, treatment with hyperthermia in mice resulted in an increase in serum IL-6 levels (48). However, we clearly demonstrated the suppressive effect of hyperthermia on IL-6 expression in mouse macrophages and MEF cells (Fig. 1). This result was further confirmed by the finding that LPS-induced IL-6 expression in ATF3-null MEF cells was not affected by hyperthermia (Fig. 3).

Because HSF1 is required for maximal expression of IL-6, partially opening the chromatin structure of the IL-6 promoter in spleen cells and MEF cells (29), we wondered whether the serum level of IL-6 increased or decreased in mice deficient in HSF1. We found that serum IL-6 levels and IL-6 mRNA levels in many tissues were extremely high in HSF1-null mice compared with wild-type mice (Fig. 5, Supplemental Fig. 5). These findings indicate for the first time that HSF1 plays a major role in the negative regulation of IL-6 expression in mice.

It is clear that heat pretreatment inhibited the LPS-induced expression of ∼86 genes (86%) (Fig. 4). These genes were divided into an ATF3-independent group containing eight genes, including TNFα and IL-1β, whose promoters may be directly inhibited by HSF1 (26, 27), and an ATF3-dependent group containing 16 genes in addition to IL-6, the expression of half of which is not maximally induced by LPS treatment in the absence of HSF1. The NOS2, ICAM1, and IL-6 promoters are bound by HSF1, and their chromatin structure is more open in the presence of HSF1. Thus, HSF1 affects chromatin status directly and also inhibits the expression of inflammatory genes directly by activating ATF3 (Supplemental Fig. 8).

HSF1 was originally identified as a transcriptional activator of heat shock proteins, and a lot of attention has focused on its role in protecting cells from lethal thermal stress (16–19, 49, 50). However, most cells in the body do not encounter such extreme temperatures (i.e., >42°C). Rather, cells are repeatedly exposed to febrile-range temperatures, which are usually elicited by exercise, feeding, infection, and disease (1, 2). In this study, we showed for the first time that HSF1 induces the expression of another transcription factor, ATF3, to regulate a unique set of genes, such as inflammatory genes. Inflammatory cytokines support the invasion of inflammatory cells and promote tissue repair; however, they are detrimental at excessively high levels. This novel negative regulatory loop may inhibit an excessive febrile response and injuries to tissues.

We thank Dr. S. Calderwood, G. Santro, and S. Kitajima for helpful discussions and Dr. K. Miyake for reagents.

Disclosures The authors have no financial conflicts of interest.